Ceph

This could be used to figure out if a monitor is up, and if it is what's its excuse for not being in the quorum.

We could either try to reuse the same mechanism used for 'mon_status' as is used by the admin socket. If we want something rather more specific, we could simply add a new message with the purpose of acting as a heartbeat sort of mechanism between clients and monitors, but for this particular case we probably don't need such approach.

We should assess what are the proper requisites for this and decide upon what we would like to obtain from such mechanism.

how about:

- new librados call ping_mon(...) that takes a timeout
- use aio completion api? or make it synchronous? aio is better i think.
- monclient needs new ping method
- track outstanding pings internally
- use separate msgr from the main session to avoid fighting with session hunting
- send MPing
- json in reply payload
- monitor
- reply to MPing with another MPing with _mon_status json in the data payload